Polimetric Postmortem

I wrote a while back that I thought there was a realistic chance that Barack Obama could throw the Prop. 8 election over to the "yes" side. That post, by the way, has become the most popular post I've ever written -- it's received more than 500 hits in less than 24 hours.

I've come to believe that is mostly true -- at least, I'm convinced that this effect happened, although I do not think that, by itself, it was the margin of victory. I do think that without it, Prop. 8 would have failed -- it was an indispensible part of the victory strategy.

Now, while I never pursued graduate-level studies in polimetrics, I'm pretty sure that as the political scientists pore over the data and do their statistical regression analyses, this will be borne out. What you need to understand about polimetrics is that it is based on the idea of cleavages -- finding differences between people that are thought to be significant, and seeing if those differences track along with voting behavior.

Certainly, all of us are individuals with our own opinions and thoughts. But we're also part of various groups and our way of looking at the world is influenced by our membership in, or lack of membership in, certain categories of people. Polimetric analysis seeks to quantify the effect of those differences. It is from these kinds of studies of voting behavior that we trace back our understanding of what it means to capture "the Black vote," for instance, or "the women's vote."

So when I found, via Obsidian Wings, the beginnings of polimetric analysis of the vote, it was where I gravitated to figure out how a state as traditionally liberal as California could have behaved as it did, approving (albeit narrowly) what looks like targeted bigotry of gay people in the political arena. The trackback goes ultimately to the CNN exit poll on Prop. 8. So now, let's dig in to the numbers.

First of all, the total votes were 5,387,939 for, and 4,883,460 against. The margin of victory was 504,479 votes.

There was effectively no gender cleavage. There were trends -- the older a voter is, the more likely the voter is to vote "yes," and the more education a voter has, the more likely the voter is to vote "no." Interestingly, there was a bulge in income -- very low-earning and very high-earning voters tended to vote "no," but middle-income earners tended to vote "yes." Looking at the electorate aas a whole, these things were pretty much a wash.

But there was a race cleavage* -- a big one:

Racial Group	Percent Voting Yes	Percent of Total Voters
White	49%	63%
African-American	70%	10%
Latino	53%	18%
Asian	49%	6%
Other	51%	3%

So, the ethnic group that voted for Prop. 8 in by far the largest proportion were Blacks. Working in the "yes" and "no" votes along these lines, we get the following results (rounded off to the nearest whole numbers):

Racial Group	Yes Votes	No Votes
White	3,170,781	3,300,200
African-American	718,998	308,142
Latino	979,891	868,960
Asian	301,979	314,305
Other	157,152	150,990

The marginal disparity along ethnic lines is quite pronounced. The most likely kind of voter to vote "yes" was a self-identified "conservative Repubilcan." The second-most likely kind of voter to vote "yes" was an African-American.

Now, there is little doubt that Obama boosted African-American turnout. Now, let me preface that. As chronicled on fivethirtyeight.com, "It is something of a myth that African-American voters do not turn out to vote." Once registered, African-Americans are as likely to vote as any other group of people. The difference is that they tend not to be registered in as high numbers as other groups. Now, in the primaries, African-Americans in California made up only about 7% of the total voters. In the general, election, that rose to 10%. What happened? Obama's post-primary voter registration drive is what happened.

Had voter registration remained at the levels they were in the primaries, we can surmise that there would have been, instead of the statistical figure of 1,027,140 votes cast by self-identified African-Americans, only 718,998 votes. Assuming further that those African-Americans would have kept their same 70% "Yes" pattern, that produces a result of 5,113,102 votes for Prop. 8 and 4,850,155 against. That cuts the margin of victory in half -- but it is still a victory for Prop. 8.

If we assume the same thing with Latino and Other* voters, we reduce the margin of victory even further, but not by much -- the margin of victory would drop to 227,820 out of a total of 9,316,159 votes cast, or 2.445%.

So it looks to me like Obama's voter registration drive had an effect. There is an important lesson to be learned here. That is that although African-American and Latino voters are predominantly registered as Democrats, they are nevertheless socially conservative. This should not be a surprise, particularly in this case. Obama's voter-registration drive piggybacked on a lot of church organizations -- and in a lot of African-American and Latino communities, churches are the only social organizations to speak of anyway. So you take an already religiously-centered group of people, and then cherry-pick people who are active churchgoers for your registration drive, and what you've done is to make the body of voters within that population to be very religious. And religious, in this context, means socially conservative, at least to the point that the voter's views on same-sex marriage will be informed by Biblical instruction condemning homosexuality** and therefore will be much more likely to vote "yes" on Prop. 8 or its equivalent.

In this coincidence do we see one of the political errors made by the opponents of Prop. 8 -- their calculus did not consider the effects of Obama's voter registration drive increasing social conservatism within the ranks of this segment of the electorate. This was probably the brainchild of arrogance -- the "No on 8" people mostly considered themselves liberals and progressives, and were generally attracted to Obama out of some combination of ideological sympathy with him and a desire to see an African-American President to advance race relations. But their intellectual arrogance was in assuming that everyone who was attracted to Obama would be attracted to him for the same reasons and therefore behave uniformly in the voting booth. This, it turns out, was a mistake.

I also do not think that the "Yes on 8" people can take a lot of credit for this. Most of them were the sorts of people you'd expect to be McCain voters. I do think they knew what they were doing when they, too, tapped into churches to advance their cause. And since they were focused on the ballot measure, they probably did not address the Presidential election all that much. My suspicion is that the conservatizing of the Black vote was not something that they had calculated on happening -- but which, of course, they were happy to benefit from.

But all of this only half (well, really about 55%) answers the question. Where did the other 227,820 "yes" votes come from? Here, statistics fail me. I can only point to several factors that I think were either mistakes or at least unsolved problems with the "no" campaign:

1. When some initial polls came out showing the proposition failing decisively, the "no on 8" leaders got complacent. They did not fund-raise they way they should have, they did not recruit good pepole they way they should have.
2. Many of the talented political organizers and strategists they could have drawn on were likely sucked up by the Obama campaign, and probably shipped out of state.
3. Obama himself always opposed same-sex marriage.
4. Advertisements for "no" were fewer and more concentrated in urban areas. The leaders of the campaign did not apparently think that "no" would have traction in the suburbs or exurbs. On a related and not inconsisent note, they may not think about non-urban places all that often, or all that accurately.
   
5. The "no" pitch is a little bit complicated and abstract to explain if you're afraid to show the voters a gay couple in a picture. Very few advertisements showed happy gay couples. Very few advertisements mentioned gay people at all, probably for fear that the image of a same-sex couple would shock viewers and create a backfire of support for "Yes." But this is what the controversy was about, and the best ads I saw -- very late in the game -- did show gay couples, with kids.
   
6. The "yes" pitch got submerged in a bunch of misinformation that the "no" crowd never found a way to counteract. The "my kids will learn about gay marriage at school" pitch for "yes" never was met with the obvious response: "Not if you're paying attention to what your kids are diong at school." This is true because in California, any parent can pull any kid out of sex education in a district that even has it in the first place, with no academic penalty. Yes, that's a complicated concept to explain in 30 seconds. But no one even tried to make that argument.
7. The "Yes on 8" posters had a neat little graphic of a family of stick figures and high-contrast selection of blue-on-yellow images. They stand out, and stick in the eye. The "No on 8" posters had white and dark red letters on a dark blue background -- the contrast was too low, and there were no images, just words. It was simply less appealing; no one thought to find a good visual icon for the "No" cause.
8. Because "Yes on 8" tapped into churches, there was a huge network of volunteers all around the state. I saw people putting "Yes on 8" signs up all over the place on Election Day. Nothing remotely like that for "No." Simply put, the "Yes" side out-spent, out-organized, and out-strategized the "No" side.
   

In some combination of the above, we can figure an extra 227,820 votes. And that is what made the difference politically.

Tomorrow, I'll muse a little bit about what those of us who were on the "No" side should do next. And what they should not do.


* Sorry, Bob, there was no separate category for Indians (or Indian-Americans, if you prefer). I didn't make up the exit poll.

** This is not the place to comment on the prevalent ignorance of other Biblical prohibitions on marriage and sex, like, say, divorce. I can rail against the hypocrisy of contemporary religious institutions elsewhere.